[Elecraft] separating RFI from band noise, PX3

[Jim Brown]

On Fri,3/25/2016 4:29 PM, Greg Troxel wrote:
> Jim Brown <jim at audiosystemsgroup.com> writes:
>
>> Yet another example of a switch-mode power supply.  Here's a
>> preliminary version of k9yc.com/KillingReceiveNoise.pdf  that might
>> help.
> I see you mention a 10 dB increase in levels between band closed and
> band open, with the notion that if you don't see that, you have too much
> local noise.  That makes a lot of sense for bands like 15m.  I don't
> quite follow how one can use that rule of thumb on 6m.

Most noise sources get weaker with increasing frequency, so there tends 
to be less noise on 10M and 6M because there's so there's much less to 
propagate.

> For 80m, presumably it should be quieter in the day.

Yes, and on 160M too.

> I wonder if it's possible to have any quantitative norms.  For example,
> looking at the waterfall on 30m with a 50 kHz span, and ref of -110 dBm,
> I'm seeing a lot of black pixels with a fair number of dark blue.

Most spectrum analyzers are best used by setting  the threshold at the 
noise floor, and with averaging set to the maximum for the amplitude 
display. This will cause random noise to average out, making signals and 
correlated noise stand out.  Do NOT use waterfall averaging.  Also, use 
reasonable settings for SCALE -- I use 24 dB for general operating to 
find signals on a dead band, etc. and 42 dB for contests where many 
stations are running high power into big antennas. The only time I use a 
wider scale than that is when I want to look at sidebands of a 
transmitter working into a dummy load.

>   There
> are a few solid lines that are obviously interference.  And I see some
> short-duration broadband pulses (horizontal lines).  So I think I have
> some issues, while others might find that better than usual.
>
> Clearly some noise sources are obvious on the panadaptor.  But most of
> what I'm seeing is not obvious.  In the end I suspect that trying to
> decide if I have local noise by making measurements is not going to work
> or be all that useful.  (Certainly measuring with power off makes sense,
> and the open-vs-closed band levels is something I'll probably try to
> really measure.)

Don't view this as a "measurement," but rather as a graphic view of your 
RF environment.  Also, the settings noted above can help.

> Also when using the PX3, I realize that the levels can be interpreted in
> two ways.  For signals narrower than the bin size, it seems that one
> should read the level as the power in the bin and hence the signal.  But
> for broad noise that is much bigger than bins, I think one should be
> thinking in terms of dBm/Hz.  In other words, choosing a 5x narrower
> span doesn't change the level of a carrier, but it lowers broadband
> noise 5x.  So referring to "-100 dBm" seems to require giving the span
> (or really the bin size, if one is comparing to non-Elecraft panadaptors).

Yes, this is generally true, and the P3 can be set to automatically 
change the scale to compensate when changing the bin size (scan width). 
I don't remember if the PX3 firmware has that feature. But we don't care 
all that much about absolute levels unless we're measuring something, 
and when chasing noise, we're usually not measuring (except, of course, 
for seeing how much we've suppressed a noise source by applying a fix). 
But none of this logic applies -- we don't care about absolute levels, 
we only want to know if our fix helped and how much.

73, Jim K9YC